%0 Journal Article %1 bernstein2020distance %A Bernstein, Jeremy %A Vahdat, Arash %A Yue, Yisong %A Liu, Ming-Yu %D 2020 %K generalization noise readings stable %T On the distance between two neural networks and the stability of learning %U http://arxiv.org/abs/2002.03432 %X This paper relates parameter distance to gradient breakdown for a broad class of nonlinear compositional functions. The analysis leads to a new distance function called deep relative trust and a descent lemma for neural networks. Since the resulting learning rule seems not to require learning rate grid search, it may unlock a simpler workflow for training deeper and more complex neural networks. Please find the Python code used in this paper here: https://github.com/jxbz/fromage.

@article{bernstein2020distance, abstract = {This paper relates parameter distance to gradient breakdown for a broad class of nonlinear compositional functions. The analysis leads to a new distance function called deep relative trust and a descent lemma for neural networks. Since the resulting learning rule seems not to require learning rate grid search, it may unlock a simpler workflow for training deeper and more complex neural networks. Please find the Python code used in this paper here: https://github.com/jxbz/fromage.}, added-at = {2020-06-16T15:29:23.000+0200}, author = {Bernstein, Jeremy and Vahdat, Arash and Yue, Yisong and Liu, Ming-Yu}, biburl = {https://www.bibsonomy.org/bibtex/2667d4b1e96e962b42f10b23d2192d38f/kirk86}, description = {[2002.03432] On the distance between two neural networks and the stability of learning}, interhash = {fc4f780396e903b23a0791f2b00796dc}, intrahash = {667d4b1e96e962b42f10b23d2192d38f}, keywords = {generalization noise readings stable}, note = {cite arxiv:2002.03432}, timestamp = {2020-06-16T15:29:23.000+0200}, title = {On the distance between two neural networks and the stability of learning}, url = {http://arxiv.org/abs/2002.03432}, year = 2020 }